Equity in IDEA: Contents of the Final Rule Presenters: Ruth Ryder, Michael Gross, Richelle Davis

Agenda Part I Significant Disproportionality Part II Equity in IDEA Final Rule Overview Part III Standard Methodology Part IV Data Reporting Part V Questions

Part I SIGNIFICANT DISPROPORTIONALITY

Racial and Ethnic Disparities IDEA 618 State-reported data (LEA MOE/CEIS Collection): Roughly 2-3% of districts identified each year as having significant disproportionality (fewer than 500) 2013-2014 school year: 75% of identified LEAs were in 8 States 25 States identified no LEAs with significant disproportionality Of States that identified LEAs, 10 only identified in one category of analysis (i.e., either identification, placement, or discipline) Only four States and one entity identified LEAs in all three categories of analysis

Racial and Ethnic Disparities IDEA 618 State-reported data: Black or African American students are twice as likely to be identified as having an emotional disturbance Black or African American students are over two times as likely to be identified as having an intellectual disability Black or African American students are 1.4 times as likely to receive services in separate settings

GAO Report: February 2013 Findings: 2% of LEAs in 2010-2011 were identified with significant disproportionality “the discretion that states have in defining significant disproportionality has resulted in a wide range of definitions that provides no assurance that the problem is being appropriately identified across the nation.”

Notice of Proposed Rulemaking Notice of Proposed Rulemaking (NPRM) was published on March 2, 2016 and was available for 75 days Received 316 comments from SEAs and LEAs Parent Advocates Researchers Associations and organizations Analyzed comments and prepared the final rule

Part II EQUITY IN IDEA FINAL RULE

Final Rule Final rule package includes An Executive Summary (helpful overview) Effective Date of these Regulations Analysis of Comments and Changes (detailed discussion of comments and our response) Regulatory Impact Analysis AND the final rule (the last 10-12 pages)

Effective Dates Final rule is effective 30 days from publication in the Federal Register BUT States have 18 months to prepare, work with their State Advisory Panel and make decisions regarding their methodology States must comply by July 1, 2018.

Methodology Statute requires States to annually collect and examine data to determine whether significant disproportionality based on race or ethnicity is occurring in the State and LEAs of the State with respect to: Identification of children as children with disabilities, including identification as children with particular impairments Placement of children in particular educational settings; and Incidence, duration, and type of disciplinary actions, including suspensions and expulsions

Methodology Final rule requires States to use a standard methodology to determine if there is significant disproportionality by examining data using a risk ratio or alternate risk ratio analysis. As part of the standard methodology, States must develop, based on advice from stakeholders (including State Advisory Panels): a reasonable risk ratio threshold, a reasonable minimum cell size a reasonable minimum n-size

Rebuttable Presumption The final regulations establish a rebuttable presumption that a minimum cell size (numerator or racial/ethnic group being analyzed) of no greater than 10 and a minimum n-size (denominator or comparison group) of no greater than 30 are reasonable

Standard Methodology Flexibilities States have the flexibility to: Consecutive Years: Use up to 3 years of data to identify an LEA with significant disproportionality Reasonable Progress: Not identify LEAs if they are demonstrating reasonable progress in lowering the applicable risk ratios in each of the two prior consecutive years

Review and Revision of Policies and Procedures A State is required to: Provide for the review of policies, practices, and procedures to ensure they comply with the IDEA LEAs are required to: Publicly report on revisions consistent with Family Education Rights and Privacy Act (FERPA)

Comprehensive CEIS LEAs identified with significant disproportionality: Are allowed to use comprehensive CEIS to serve students, ages 3 through grade 12, with and without disabilities Are required to identify and address the factors that may contribute to the significant disproportionality

Implementation OSEP will support implementation of the new requirements set forth by the Equity in IDEA final rule by: Providing technical assistance and developing guidance to support States as they prepare to implement the new requirements, see https://www.osepideasthatwork.org/federal-resources-stakeholders/disproportionality-and-equity   Working with TA Centers such as the IDEA Data Center to support data analysis and implementation

Part III STANDARD METHODOLOGY

Definitions Made Easy Risk(risk index): Risk tells us how likely a certain outcome is (i.e. being identified as having a disability) Comparison group: All other races Risk ratio: The risk ratio tells us how the risk for one racial/ethnic group compares to the risk for a comparison group Minimum cell size: Risk numerator Minimum n-size: Risk denominator Alternate risk ratio: Uses the district level risk for racial/ethnic group in the numerator and the state level risk for the comparison group. Used if the comparison group does not meet the minimum cell or n-size

Standard Methodology RISK A proportion expressing likelihood. Example: 40 Hispanic children identified 200 total Hispanic children in LEA Risk of Hispanic child identified as child with disability = 40/200 or 20%.

Standard Methodology RISK RATIO A comparison of risks: likelihood of outcome for one group vs. outcome for all others in the LEA Example: 40 Hispanic children identified out of 200 total Hispanic children in LEA 200 of other children identified out of All 2,000 other children in LEA Risk ratio: 2.0 (40/200) / (200/2000) = 0.2 / 0.1 = 2.0

RISK RATIO 2.0 = 2x as likely 3.0 = 3x as likely Etc. Standard Methodology RISK RATIO 2.0 = 2x as likely 3.0 = 3x as likely Etc.

Standard Methodology RISK RATIO THRESHOLD What is that significant disproportionality? A risk ratio > the risk ratio threshold = significant disproportionality State must set risk ratio threshold in consultation with stakeholders, including SAP.

Standard Methodology RISK RATIO THRESHOLD How many? 14 One for each category of analysis May set different threshold for each, as reasonable.

CATEGORIES OF ANALYSIS: Standard Methodology CATEGORIES OF ANALYSIS: The identification of children ages 3 through 21 as children with disabilities; The identification of children ages 3 through 21 as children with the following impairments: Intellectual disabilities; Specific learning disabilities; Emotional disturbance; Speech or language impairments; Other health impairments; and Autism. Placements of children with disabilities ages 6 through 21, inside a regular class less than 40 percent of the day; Placements of children with disabilities ages 6 through 21, inside separate schools and residential facilities, not including homebound or hospital settings, correctional facilities, or private schools; For children with disabilities ages 3 through 21, out-of-school suspensions and expulsions of 10 days or fewer; For children with disabilities ages 3 through 21, out-of-school suspensions and expulsions of more than 10 days; For children with disabilities ages 3 through 21, in-school suspensions of 10 days or fewer; For children with disabilities ages 3 through 21, in-school suspensions of more than 10 days; and For children with disabilities ages 3 through 21, disciplinary removals in total, including in-school and out-of-school suspensions, expulsions, removals by school personnel to an interim alternative education setting, and removals by a hearing officer. (34 C.F.R. §300.647(b)(3) and (4).)

Standard Methodology Categories of Analysis are applied to each of 7 racial or ethnic groups: Hispanic/Latino of any race, and for individuals who are non-Hispanic/Latino only; American Indian or Alaska Native; Asian; Black or African American;  Native Hawaiian or Other Pacific Islander;  White; and  Two or more races. (34 C.F.R. §300.647(b)(2).)

Standard Methodology “SETTINGS” IN STANDARD METHODOLOGY State must set (in consultation with SAP etc.) Risk Ratio Threshold and what else? Reasonable minimum cell size Reasonable minimum n-size Period of SD determination (Up to 3 years, optional) Definition of Reasonable Progress (Optional)

Standard Methodology CELL SIZES AND N-SIZES Example: 40 Hispanic children identified out of [cell size] 200 total Hispanic children in LEA [n-size] 200 of other children identified out of [cell size] All 2,000 other children in LEA [n-size] Risk ratio: 2.0 (40/200) / (200/2000) = 0.2 / 0.1 = 2.0

Standard Methodology WHY MINIMUM CELL SIZES AND N-SIZES? Risk ratios can produce unreliable or volatile numbers when applied to small populations. Determinations of significant disproportionality should not turn on small demographic changes.

Standard Methodology WHY MINIMUM CELL SIZES AND N-SIZES? Example : In a small LEA, the ratio threshold for Native students identified as CWD = 3.0 4 Native children identified out of [cell size] 8 total Native children in LEA [n-size] 10 children identified out of [cell size] All 50 other children in LEA [n-size] Risk ratio: (4/8) / (10/50) = 0.5 / 0.2 = 2.5

Standard Methodology WHY MINIMUM CELL SIZES AND N-SIZES? Example cont’d: Next year, 2 Native students with disabilities moved into the LEA: 6 Native children identified out of [cell size] 10 total Native children in LEA [n-size] 10 children identified out of [cell size] All 50 other children in LEA [n-size] Risk ratio: (6/10) / (10/50) = 0.6 / 0.2 = 3.0 the SD threshold

Standard Methodology MINIMUM CELL SIZES AND N-SIZES Must be reasonable. Minimum cell sizes <= 10 Minimum n-sizes <= 30 Are presumptively reasonable

Standard Methodology ALTERNATE RISK RATIO A comparison of risks: likelihood of outcome for one group vs. outcome for all others in the State Because sometimes the comparison group won’t meet the minimum cell or n-size.

Standard Methodology ALTERNATE RISK RATIO Example: Minimum cell size = 5 Minimum n-size = 30 490 out of 500 students in the LEA are Native American / Alaska Native Number of other students in comparison group = 10, so: 70 Native American / Alaska Native children identified as CWD out of 490 total NA / AN children in the LEA 520,000 children identified as CWD out of 3,640,000 all other children in the State Alternate risk ratio: (70 / 490) / (520,000 / 3,640,000) = 1.43 / 1.43 = 1.0

Standard Methodology “CONSECUTIVE YEARS” FLEXIBILTY or USE “MULTIPLE YEARS OF DATA” States may choose to make determinations of significant disproportionality only after multiple years (up to three) Because risk ratios can be volatile and because systematic change can take time.

Standard Methodology “CONSECUTIVE YEARS” FLEXIBILTY or USE “MULTIPLE YEARS OF DATA” Example: In school year 2019-2020, a State has set a risk ratio threshold for identification of 3.0 and requires an LEA to exceed the threshold for three consecutive years: Only LEA 2 will be determined to have significant disproportionality in identification, despite the risk ratio of 3.3 for LEA 1 in 2016-17.   2015-16 2016-17 2017-18 LEA 1 2.7 3.3 2.6 LEA 2 3.1

“REASONABLE PROGRESS” FLEXIBILITY Standard Methodology “REASONABLE PROGRESS” FLEXIBILITY Optional If LEA above risk ratio threshold but lowering risk ratio for the two prior consecutive years, State need not find significant disproportionality. Specific details of how much risk ratio must be lowered is determined by State in consultation with stakeholders, including SAP. Given the time it takes to make systematic change, why interrupt something that is working.

“REASONABLE PROGRESS” FLEXIBILITY Standard Methodology “REASONABLE PROGRESS” FLEXIBILITY Example: State has set a risk ratio threshold for identification. State has defined “reasonable progress” to mean a year-to-year decline in risk ratio of 0.5. In school year 2021-2022, the State need not find significant disproportionality for identification in LEA 1.   2018-19 2019-20 2020-21 LEA 1 4.9 4.3 3.6 LEA 2

Part IV DATA REPORTING

Data Collection and Reporting OSEP is still working through the full implications of data collection SEAs already provide the majority of the necessary data/information to ED It is anticipated that the MOE/CEIS collection will have some edits in the coming years An additional collection vehicle is being discussed to capture the required collection pieces addressed in the Final Rule

File Specifications Used We anticipate SEAs and LEAs will use counts from these already collected EDFacts file specifications to calculate risk ratios: File Description C002 Children with Disabilities (IDEA) School Age C089 Children with Disabilities (IDEA) Early Childhood C006 Children with Disabilities (IDEA) Suspensions/Expulsions C143 Children with Disabilities (IDEA) Total Disciplinary Removals C052 Membership

Educational environments Total disciplinary removals How the files are used   Identification Educational environments Suspension/ expulsion Total disciplinary removals Risk numerator / cell size C002 C006 C143 Risk denominator / n-size C052 C089 Note: OSEP will provide further clarification around the appropriate risk denominator for 3-5 year olds in the coming months

File Specification Example Risk Numerator Must have minimum cell size of 10 Risk Ratio = C002: Number of children from racial or ethnic group in a disability category C052: Number of enrolled children from racial or ethnic group ÷ C002: Number of children from all other racial or ethnic group in a disability category C052: Number of enrolled children from all other racial or ethnic groups Risk Denominator Must have minimum N- size of 30

SSS-IDEA User Guide Updated SSS-IDEA User Guide has been updated to identify the requested pieces of a State’s definition of significant disproportionality The definition should include the following elements, as appropriate: The calculation method(s) being used (i.e. risk ratio, weighted risk ratio, eformula, etc.); Any minimum cell- or n-sizes (i.e. risk numerator and/or risk denominator); The number of years of data used in the calculation; and The threshold at which significant disproportionality is identified.

Data Quality Plug Data quality is increasingly important, especially at LEA and school level More and more of the 618 data is being used for high stakes decisions Building in more sophisticated edit checks Increases the role of strong meta-data and meta-data to collected data checks State data quality has been improving over the past several years Use lessons learned at SEA-level to continue improving LEA and school data

Part V QUESTIONS Please send questions for us to address in future guidance at significantdisproportionalityrule@ed.gov